On this tutorial, we take an in depth, sensible strategy to exploring NVIDIA’s KVPress and understanding the way it could make long-context language mannequin inference extra environment friendly. We start by establishing the total atmosphere, putting in the required libraries, loading a compact Instruct mannequin, and getting ready a easy workflow that runs in Colab whereas nonetheless demonstrating the actual worth of KV cache compression. As we transfer by way of implementation, we create an artificial long-context corpus, outline focused extraction questions, and run a number of inference experiments to instantly evaluate normal technology with totally different KVPress methods. On the finish of the tutorial, we can have constructed a stronger instinct for the way long-context optimization works in apply, how totally different press strategies have an effect on efficiency, and the way this type of workflow might be tailored for real-world retrieval, doc evaluation, and memory-sensitive LLM functions.
import os, sys, subprocess, textwrap, time, gc, json, math, random, warnings, examine
warnings.filterwarnings("ignore")
def run(cmd):
print("n[RUN]", " ".be part of(cmd))
subprocess.check_call(cmd)
run([sys.executable, "-m", "pip", "install", "-q", "--upgrade", "pip"])
run([sys.executable, "-m", "pip", "install", "-q", "torch", "transformers", "accelerate", "bitsandbytes", "sentencepiece", "kvpress==0.4.0"])
attempt:
from google.colab import userdata
hf_token = userdata.get("HF_TOKEN")
besides Exception:
hf_token = os.environ.get("HF_TOKEN", "")
if not hf_token:
attempt:
import getpass
hf_token = getpass.getpass("Enter your Hugging Face token (leave empty if model is public and accessible): ").strip()
besides Exception:
hf_token = ""
if hf_token:
os.environ["HF_TOKEN"] = hf_token
os.environ["HUGGINGFACEHUB_API_TOKEN"] = hf_token
import torch
import transformers
import kvpress
from transformers import pipeline, BitsAndBytesConfig
from kvpress import ExpectedAttentionPress, KnormPress
print("Python:", sys.model.break up()[0])
print("Torch:", torch.__version__)
print("Transformers:", transformers.__version__)
print("CUDA available:", torch.cuda.is_available())
if torch.cuda.is_available():
print("GPU:", torch.cuda.get_device_name(0))
MODEL_ID = "Qwen/Qwen2.5-1.5B-Instruct"
MAX_NEW_TOKENS = 96
SEED = 42
random.seed(SEED)
torch.manual_seed(SEED)We arrange the Colab atmosphere and set up all required libraries to run the KVPress workflow efficiently. We securely acquire the Hugging Face token, set atmosphere variables, and import the core modules wanted for mannequin loading, pipeline execution, and compression experiments. We additionally print the runtime and {hardware} particulars so we clearly perceive the setup by which we carry out the tutorial.
if torch.cuda.is_available():
torch.cuda.empty_cache()
quantization_config = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_compute_dtype=torch.float16,
bnb_4bit_quant_type="nf4",
bnb_4bit_use_double_quant=True,
)
pipe = pipeline(
"kv-press-text-generation",
mannequin=MODEL_ID,
device_map="auto",
token=hf_token if hf_token else None,
model_kwargs={
"quantization_config": quantization_config,
"attn_implementation": "sdpa",
},
)
else:
pipe = pipeline(
"kv-press-text-generation",
mannequin=MODEL_ID,
device_map="auto",
torch_dtype=torch.float32,
token=hf_token if hf_token else None,
model_kwargs={
"attn_implementation": "sdpa",
},
)
def cuda_mem():
if not torch.cuda.is_available():
return {"allocated_gb": None, "reserved_gb": None, "peak_gb": None}
return {
"allocated_gb": spherical(torch.cuda.memory_allocated() / 1024**3, 3),
"reserved_gb": spherical(torch.cuda.memory_reserved() / 1024**3, 3),
"peak_gb": spherical(torch.cuda.max_memory_allocated() / 1024**3, 3),
}
def reset_peak():
if torch.cuda.is_available():
torch.cuda.reset_peak_memory_stats()
def extract_answer(x):
if isinstance(x, checklist) and len(x) > 0:
x = x[0]
if isinstance(x, dict):
for ok in ["answer", "generated_text", "text", "output_text"]:
if ok in x:
return x[k]
return json.dumps(x, indent=2, ensure_ascii=False)
return str(x)
def generate_once(context, query, press=None, label="run"):
gc.acquire()
if torch.cuda.is_available():
torch.cuda.empty_cache()
reset_peak()
begin = time.time()
out = pipe(
context,
query=query,
press=press,
max_new_tokens=MAX_NEW_TOKENS,
do_sample=False,
temperature=None,
return_full_text=False,
)
elapsed = time.time() - begin
reply = extract_answer(out)
stats = cuda_mem()
end result = {
"label": label,
"elapsed_sec": spherical(elapsed, 2),
"allocated_gb": stats["allocated_gb"],
"reserved_gb": stats["reserved_gb"],
"peak_gb": stats["peak_gb"],
"answer": reply.strip(),
}
return end resultWe initialize the kv-press-text-generation pipeline and configure it otherwise relying on whether or not GPU help is out there. We outline the helper capabilities that measure CUDA reminiscence utilization, reset peak reminiscence, extract solutions from mannequin outputs, and run a single technology go cleanly. This half gives the reusable execution logic that powers the remainder of the tutorial and permits us to match baseline inference with KV cache compression.
company_records = [
{"company": "Arcturus Dynamics", "hq": "Bengaluru", "founded": 2017, "focus": "warehouse robotics"},
{"company": "BlueMesa Energy", "hq": "Muscat", "founded": 2014, "focus": "grid analytics"},
{"company": "CinderPeak Health", "hq": "Pune", "founded": 2019, "focus": "clinical imaging AI"},
{"company": "DeltaForge Marine", "hq": "Kochi", "founded": 2012, "focus": "autonomous vessel telemetry"},
{"company": "EonCircuit Labs", "hq": "Hyderabad", "founded": 2020, "focus": "edge silicon tooling"},
{"company": "Frostline Aero", "hq": "Jaipur", "founded": 2016, "focus": "drone inspection"},
]
needle_facts = [
"PROJECT NEEDLE 1: The internal codename for the confidential pilot program is SAFFRON-17.",
"PROJECT NEEDLE 2: The audit escalation owner is Meera Vashisht.",
"PROJECT NEEDLE 3: The approved deployment region for the first production rollout is Oman North.",
"PROJECT NEEDLE 4: The emergency rollback phrase is amber lantern.",
"PROJECT NEEDLE 5: The signed commercial start date is 17 September 2026.",
]
background_block = """
Lengthy-context techniques usually comprise repeated operational notes, historic information, coverage sections, and noisy retrieval artifacts.
The aim of this demo is to create a realistically lengthy immediate the place only some particulars matter for downstream answering.
KV cache compression reduces reminiscence utilization by pruning cached key-value pairs whereas preserving reply high quality.
"""
policy_block = """
Operational coverage abstract:
1. Security overrides throughput when sensor confidence falls under threshold.
2. Logs ought to protect area, timestamp, machine class, and operator approval state.
3. Subject trials might comprise duplicated annexes, OCR-style artifacts, and repeated compliance summaries.
4. A superb long-context mannequin should ignore irrelevant repetition and retrieve the precise particulars that matter.
"""
records_text = []
for i in vary(120):
rec = company_records[i % len(company_records)]
records_text.append(
f"Record {i+1}: {rec['company']} is headquartered in {rec['hq']}, founded in {rec['founded']}, and focuses on {rec['focus']}. "
f"Quarterly memo {i+1}: retention remained stable, operator training progressed, and the compliance appendix was reattached for review."
)
needle_insert_positions = {18, 41, 73, 96, 111}
full_corpus = []
for i, para in enumerate(records_text):
full_corpus.append(background_block.strip())
full_corpus.append(policy_block.strip())
full_corpus.append(para)
if i in needle_insert_positions:
full_corpus.append(needle_facts[len([x for x in needle_insert_positions if x <= i]) - 1])We create an artificial long-context dataset to check the KVPress system in a managed but reasonable manner. We outline firm information, insert vital hidden info at totally different positions, and blend them with repeated background and coverage blocks, making the immediate lengthy and noisy. This helps us simulate the context by which memory-efficient inference issues and the mannequin should retrieve solely the actually related particulars.
context = "nn".be part of(full_corpus)
query = textwrap.dedent("""
Reply utilizing solely the offered context.
Give a compact JSON object with precisely these keys:
commercial_start_date
deployment_region
audit_owner
rollback_phrase
pilot_codename
""").strip()
print("nContext characters:", len(context))
print("Approx words:", len(context.break up()))
experiments = []
baseline = generate_once(context, query, press=None, label="baseline_no_compression")
experiments.append(baseline)
presses = [
("expected_attention_0.7", ExpectedAttentionPress(compression_ratio=0.7)),
("expected_attention_0.5", ExpectedAttentionPress(compression_ratio=0.5)),
("knorm_0.5", KnormPress(compression_ratio=0.5)),
]
for label, press in presses:
attempt:
end result = generate_once(context, query, press=press, label=label)
experiments.append(end result)
besides Exception as e:
experiments.append({
"label": label,
"elapsed_sec": None,
"allocated_gb": None,
"reserved_gb": None,
"peak_gb": None,
"answer": f"FAILED: {type(e).__name__}: {e}"
})
attempt:
from kvpress import DecodingPress
sig = examine.signature(DecodingPress)
kwargs = {"base_press": KnormPress()}
if "compression_interval" in sig.parameters:
kwargs["compression_interval"] = 10
elif "compression_steps" in sig.parameters:
kwargs["compression_steps"] = 10
if "target_size" in sig.parameters:
kwargs["target_size"] = 512
elif "token_buffer_size" in sig.parameters:
kwargs["token_buffer_size"] = 512
if "hidden_states_buffer_size" in sig.parameters:
kwargs["hidden_states_buffer_size"] = 0
decoding_press = DecodingPress(**kwargs)
decoding_result = generate_once(context, query, press=decoding_press, label="decoding_knorm")
experiments.append(decoding_result)
besides Exception as e:
experiments.append({
"label": "decoding_knorm",
"elapsed_sec": None,
"allocated_gb": None,
"reserved_gb": None,
"peak_gb": None,
"answer": f"SKIPPED_OR_FAILED: {type(e).__name__}: {e}"
})We assemble the ultimate context, outline the structured extraction query, and launch the core set of inference experiments. We first run the baseline with out compression, then apply a number of press methods to look at how totally different compression ratios have an effect on the outcomes. We additionally conduct a decoding-oriented compression experiment, which extends the tutorial past prefilling and gives a broader view of the KVPress framework.
print("n" + "=" * 120)
print("RESULTS")
print("=" * 120)
for r in experiments:
print(f"n[{r['label']}]")
print("elapsed_sec:", r["elapsed_sec"])
print("allocated_gb:", r["allocated_gb"])
print("reserved_gb:", r["reserved_gb"])
print("peak_gb:", r["peak_gb"])
print("answer:")
print(r["answer"])
print("n" + "=" * 120)
print("SIMPLE SUMMARY")
print("=" * 120)
def safe_float(x):
attempt:
return float(x)
besides Exception:
return None
base_peak = safe_float(baseline["peak_gb"]) if baseline.get("peak_gb") just isn't None else None
base_time = safe_float(baseline["elapsed_sec"]) if baseline.get("elapsed_sec") just isn't None else None
for r in experiments[1:]:
peak = safe_float(r["peak_gb"])
t = safe_float(r["elapsed_sec"])
peak_delta = None if base_peak is None or peak is None else spherical(base_peak - peak, 3)
time_delta = None if base_time is None or t is None else spherical(base_time - t, 2)
print({
"label": r["label"],
"peak_gb_saved_vs_baseline": peak_delta,
"time_sec_saved_vs_baseline": time_delta,
"answer_preview": r["answer"][:180].substitute("n", " ")
})
print("n" + "=" * 120)
print("OPTIONAL NEXT STEPS")
print("=" * 120)
print("1. Swap MODEL_ID to a stronger long-context instruct model that fits your GPU.")
print("2. Increase context length by duplicating records_text more times.")
print("3. Try other presses from kvpress, such as SnapKVPress, StreamingLLMPress, QFilterPress, or ChunkKVPress.")
print("4. Replace the synthetic corpus with your own long PDF/text chunks and keep the same evaluation loop.")We print all experiment outputs in a readable format and summarize the runtime and reminiscence variations relative to the baseline. We calculate easy comparability metrics to shortly see how a lot reminiscence or time every compression technique saves. We then conclude with instructed subsequent steps to increase the tutorial to stronger fashions, longer contexts, extra press strategies, and real-world doc workloads.
In conclusion, we developed a powerful sensible understanding of how NVIDIA’s KVPress can be utilized to optimize long-context inference in a practical Colab-based setting. We did greater than merely run a mannequin: we constructed an end-to-end workflow that installs the framework, masses the pipeline accurately, constructs a significant long-context enter, applies a number of compression presses, and evaluates the outcomes when it comes to reply high quality, runtime, and reminiscence habits. By evaluating baseline technology with compressed KV-cache technology, we clearly noticed the trade-offs concerned. We gained helpful instinct about when these strategies will help cut back useful resource strain with out severely harming output constancy. We additionally explored the framework’s flexibility by testing totally different press configurations and together with an non-compulsory decoding-oriented compression path, offering a broader view of how KVPress can be utilized past a single static instance.
Try the Codes and Pocket book right here. Additionally, be at liberty to comply with us on Twitter and don’t overlook to hitch our 120k+ ML SubReddit and Subscribe to our Publication. Wait! are you on telegram? now you possibly can be part of us on telegram as effectively.
Must accomplice with us for selling your GitHub Repo OR Hugging Face Web page OR Product Launch OR Webinar and so on.? Join with us
